|
|
|
|
|
|
by throw123890423
256 days ago
|
|
|
> I will say, their pricing and deployment strategy is a bit murky and unclear. Paying $1500-$10,000 per month plus usage costs? I'm assuming that it has to do with chasing and optimizing for higher value contracts and deeper-pocketed customers, hence the minimum monthly spend that they require. Yeah wait, why rent chips instead of sell them? Why wouldn't customers want to invest money in competition for cheaper inference hardware? It's not like Nvidia has a blacklist of companies that have bought chips from competitors, or anything. Now that would be crazy! That sure would make this market tough to compete in, wouldn't it. I'm so glad Nvidia is definitely not pressuring companies to not buy from competitors or anything. |
|
|
1. They’re useless for training in 2025. They were designed for training prior to LLM explosion. They’re not practical for training anymore because they rely on SRAM which is not scalable.
2. No one is going to spend the resources to optimize models to run on their SDK and hardware. Open source inference engines don’t optimize for Cerebras hardware.
Given the above two reasons, it makes a lot of sense that no one is investing in their hardware and they have switched to a cloud model selling speed as the differentiator.
It’s not always “Nvidia bad”.